You know the firm needs more money for AI. But how to ask?
Somewhere in this year's budget conversation, someone will ask what last year's AI investment returned.
You must have thought about the possible answers. Usage is up. Lawyers say they save time. There are a few nice anecdotes. More practices are experimenting, and more OCGs are asking what the firm is using AI for.
You also know that sooner or later someone will want the number: what did we spend, and what did we get back? For most firms, AI ROI is still surprisingly hard to calculate, which is awkward when you are about to ask for more money for next year.
The tempting way to prove AI ROI
I am sure you have tried to build the ROI case anyway. Estimate the hours saved, multiply by rates, extrapolate across users, and arrive at a satisfyingly large number.
The arithmetic works. The economics don't.
An hour saved isn't necessarily an hour sold. On work billed by the hour, faster can mean less revenue rather than more margin, and freed-up capacity only becomes money when something happens to it. Creating that missing link isn't impossible, but it is extremely cumbersome. I am not sure it is even worth the trouble.
Perhaps the problem is that we are looking for the return in the wrong place.
Stop promising a return on the licenses
Most of a law firm's AI budget is foundation: licenses, security, training, integrations, and increasingly the AI features arriving in software you already own. It is necessary, and I think it deserves to be defended as what it is: the cost of staying in the game. Nobody calculates the ROI of email.
The return has to come from somewhere else. It comes when lawyers start doing something different with these tools: offering a service clients couldn't get before, changing what a matter costs, or pricing work in a way that rewards speed instead of punishing it.
That is also where next year's answer to what did we get back? will eventually have to come from.
The resource your AI budget doesn't cover
This is where the budget conversation quietly stops being about technology.
Changing how legal work is delivered and priced can't be done by IT or an innovation team alone. It takes lawyers who care enough about a client problem to spend scarce time on it, who stay close enough to notice when an assumption turns out to be wrong, and who are willing to try something that may not work.
Which makes lawyer attention one of the scarcest resources in your AI budget, and one that is remarkably difficult to buy.
You can buy the licenses. You can hire the technical expertise. You can provide training, build integrations and make the tools increasingly easy to use. None of that means a partner with a full book of business will decide that changing how a piece of legal work is delivered is the best use of Tuesday afternoon.
And yet that is roughly what needs to happen if the firm wants a return beyond productivity.
So how do you get lawyers to spend time on it?
Most firms I talk to eventually arrive at some version of the same answer: hire someone.
Someone who understands AI and the business of law, who can work with partners, redesign workflows, and help practices figure out where the opportunities are. Ideally, someone who can inspire people to move as well.
I understand the instinct. I'm also fairly sure it won't work, and not only because that person is so hard to find. It rests on an assumption about how lawyers change that doesn't hold up. And the people who could actually help you are probably already inside your firm, doing something your budget doesn't see yet.
I wrote a short piece about that assumption, and about why it is good news that the hire most firms are hoping for barely exists.
Read the one-page argument: Fortunately, Unicorns Don't Exist. [Download]